There have been several attempts to completely rewrite MML (Simba's core - recreating the actual UI is a non-issue, obviously) in C, C++, and other languages, I'm sure. I think the biggest advantage would be that Pascal definitely has it's share of oddities, but nothing that's a complete showstopper. Keep in mind that Simba itself was a complete rewrite and it took quite a few people quite a lot of time to get it where it is now. To just throw all that away seems like a mistake, imo. There's nothing that awful about MML, and it supports most everything people need it to (multithreading is really the only thing that people request, but I don't think it's necessary personally), and there's nothing about MML that makes it impossible to add any features if people spend some time with it.
Anyway, I'm curious about how you worded this post I'm replying to. Do you mean someone here is actively working on a complete rewrite of MML with some .NET language? If so, I'm curious. I suppose now would be a good time for something like that with Microsoft releasing so much .NET stuff for free with plans on Linux and Mac OS X support (and just having Mono improved). However, I still think it's an extreme amount of work for not much gain.
So my point is that you should look into libMML! It doesn't "fix" all the problems you mentioned, but it would allow people to write scripts in whatever language you want (all .NET languages would work fine, including F#, and I've wanted to see a functional language being used for scripting for awhile)! I just recently pushed some more updates to libMML and pyMML the other day, which you can read about
here. I've been busy with the holidays and everything since then, but I do plan on finishing more, adding tests and documentation, etc. soon. If you want to help, I'd be more than happy to help you get involved!